Fix/29365 duplicate tool responses by abhashkumar9051 · Pull Request #29400 · google-gemini/gemini-cli

abhashkumar9051 · 2026-09-18T19:26:24Z

Summary

Fixes duplicate functionResponse messages when resuming Gemini CLI sessions with -r.

Tool results could be persisted both in toolCalls[].result and as durable user messages. During session restoration, both copies could be replayed, causing duplicate function responses for the same function call ID.

This can cause strict Gemini-compatible backends to reject the resumed request because the number of function responses does not match the number of function calls.

Details

The fix makes durable user functionResponse records the preferred source of truth during session restoration.

Changes include:

convertSessionToClientHistory() now tracks functionResponse IDs already stored in user messages.
Tool results from toolCalls[].result are only regenerated when a durable response for the same call ID does not exist.
Duplicate functionResponse parts with the same call ID are filtered during history reconstruction.
Tool-call result synchronization now stores only the response matching the current call ID.
Parallel tool-call responses remain isolated and no longer inherit responses belonging to sibling calls.

Related Issues

Fixes #29365

How to Validate

Run the relevant test suites and verify that:

Sessions containing persisted tool results can be resumed without duplicate functionResponse parts.
Previously corrupted sessions containing repeated responses can be reconstructed safely.
Parallel tool calls keep their responses associated with the correct call IDs.
The Darwin PTY orphan-FD regression test correctly detects, closes, and returns the matching file descriptor.

Validation performed:

Targeted tests: 62 tests passed
ESLint: Passed
TypeScript typecheck: Passed
Regression tests: Passed
Parallel tool-call tests: Passed
Previously failing Darwin PTY test: Fixed and re-tested successfully

Pre-Merge Checklist

Added/updated tests (if needed)
Updated the existing Darwin PTY orphan-FD cleanup test by mocking realpathSync to make PTY path resolution deterministic.
The Darwin test initially failed with Expected: 12, Received: undefined; after the test setup was corrected, it correctly identifies FD 12, closes it, and returns 12.

google-cla · 2026-09-18T19:26:29Z

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

gemini-code-assist · 2026-09-18T19:26:33Z

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses an issue where Gemini CLI sessions resumed with the -r flag would generate duplicate functionResponse messages. By ensuring that durable user messages are treated as the source of truth and filtering out redundant or incorrect tool results during history reconstruction, the fix prevents backend rejection errors caused by mismatched function call and response counts. Additionally, it improves the isolation of parallel tool calls and stabilizes existing regression tests.

Highlights

Session Restoration Fix: Implemented logic to prioritize durable user functionResponse records over toolCalls results during session restoration to prevent duplicate function responses.
Duplicate Filtering: Added tracking for functionResponse IDs to filter out duplicates during history reconstruction and ensure only the matching response is used for each tool call.
Parallel Tool Call Isolation: Updated tool-call result synchronization to ensure parallel tool calls remain isolated and do not inherit responses belonging to sibling calls.
Test Improvements: Updated the Darwin PTY orphan-FD cleanup test by mocking realpathSync to ensure deterministic path resolution.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

github-actions · 2026-09-18T19:26:38Z

📊 PR Size: size/M

Lines changed: 219
Additions: +199
Deletions: -20
Files changed: 5

gemini-code-assist

Code Review

This pull request modifies the session history conversion and recording synchronization logic to prevent replaying duplicate tool results and sibling call responses during parallel tool execution. However, the feedback highlights that the current implementation of filtering tool call results by matching ID discards multi-modal sibling parts (such as inlineData or text parts). The reviewer recommends consolidating the filtering logic into a shared utility function that preserves these sibling parts, and restoring the corresponding test cases and assertions in chatRecordingService.test.ts to ensure multi-modal tool outputs are not lost.

gemini-code-assist · 2026-09-18T19:30:32Z

+                const matchingPart = (turn.content.parts || []).find(
+                  (candidate) => candidate.functionResponse?.id === callId,
+                );
                if (
-                  JSON.stringify(tc.result) !==
-                  JSON.stringify(turn.content.parts)
+                  matchingPart &&
+                  JSON.stringify(tc.result) !== JSON.stringify([matchingPart])
                ) {
-                  tc.result = turn.content.parts || [];
+                  tc.result = [matchingPart];


By using .find() and wrapping only the single matching functionResponse part in an array, any multi-modal sibling parts (such as inlineData or text parts returned by the tool) are completely discarded during history synchronization. This breaks multi-modal tool output preservation.

Instead, we should filter the parts to keep both the matching functionResponse and any non-functionResponse sibling parts. To avoid duplication and improve maintainability, this filtering logic should be consolidated into a shared utility function.

const matchingParts = filterToolParts(turn.content.parts || [], callId); if ( matchingParts.length > 0 && JSON.stringify(tc.result) !== JSON.stringify(matchingParts) ) { tc.result = matchingParts; updated = true; }

References

When adding new functionality, such as filtering, consolidate it with existing similar logic to avoid duplication and improve maintainability.

gemini-code-assist · 2026-09-18T19:30:32Z

+                functionResponseParts.push(
+                  ...ensurePartArray(toolCall.result).filter(
+                    (part) => part.functionResponse?.id === toolCall.id,
+                  ),
+                );


Filtering the toolCall.result array to only keep parts where part.functionResponse?.id === toolCall.id will discard any multi-modal sibling parts (such as inlineData or text parts) associated with the tool call. This causes multi-modal tool outputs to be lost when resuming a session.

We should update the filter to also preserve any non-functionResponse parts. To avoid duplication and improve maintainability, this filtering logic should be consolidated into a shared utility function.

functionResponseParts.push( ...filterToolParts(ensurePartArray(toolCall.result), toolCall.id), );

References

When adding new functionality, such as filtering, consolidate it with existing similar logic to avoid duplication and improve maintainability.

gemini-code-assist · 2026-09-18T19:30:32Z

    });

-    it('should preserve multi-modal sibling parts during sync', async () => {
+    it('should sync only the matching function response during sync', async () => {


Rename the test back to reflect that it is verifying the preservation of multi-modal sibling parts.

Suggested change

it('should sync only the matching function response during sync', async () => {

it('should preserve multi-modal sibling parts during sync', async () => {

gemini-code-assist · 2026-09-18T19:30:32Z

+      expect(result).toHaveLength(1);
      expect(result[0].functionResponse!.response).toEqual({
        output: maskedSnippet,
      });
-      expect(result[1].inlineData).toBeDefined();
-      expect(result[1].inlineData!.mimeType).toBe('image/png');
    });


Since we should preserve multi-modal sibling parts (like inlineData) during history synchronization, we should restore the original assertions of this test to verify that the sibling inlineData part is not discarded.

expect(result).toHaveLength(2); expect(result[0].functionResponse!.response).toEqual({ output: maskedSnippet, }); expect(result[1].inlineData).toBeDefined(); expect(result[1].inlineData!.mimeType).toBe('image/png'); });

gemini-cli · 2026-09-26T01:45:59Z

Hi there! Thank you for your interest in contributing to Gemini CLI.

To ensure we maintain high code quality and focus on our prioritized roadmap, we only guarantee review and consideration of pull requests for issues that are explicitly labeled as 'help wanted'.

This PR will be closed in 7 days if it remains without that designation. We encourage you to find and contribute to existing 'help wanted' issues in our backlog! Thank you for your understanding.

gemini-cli · 2026-10-03T01:50:57Z

This pull request is being closed as it has been open for 14 days without a 'help wanted' designation. We encourage you to find and contribute to existing 'help wanted' issues in our backlog! Thank you for your understanding.

abhashkumar9051 requested a review from a team as a code owner September 18, 2026 19:26

github-actions Bot added the size/m A medium sized PR label Sep 18, 2026

gemini-code-assist Bot reviewed Sep 18, 2026

View reviewed changes

gemini-cli Bot added priority/p1 Important and should be addressed in the near term. area/core Issues related to User Interface, OS Support, Core Functionality labels Sep 18, 2026

abhashkumar9051 force-pushed the fix/29365-duplicate-tool-responses branch 2 times, most recently from 052fe11 to ff6559e Compare September 18, 2026 21:07

abhashkumar9051 added 2 commits September 19, 2026 02:48

issue no. google-gemini#29365 by Abhash

6c22ab0

issue no.29365 all test case passed

5dac544

abhashkumar9051 force-pushed the fix/29365-duplicate-tool-responses branch from ff6559e to 5dac544 Compare September 18, 2026 21:18

Merge branch 'main' into fix/29365-duplicate-tool-responses

a1f98da

This was referenced Sep 19, 2026

📊 AI CLI 工具社区动态日报 2026-09-19 litang9/big_model_radar#418

Open

📊 AI CLI 工具社区动态日报 2026-09-19 96loveslife/big_model_radar#512

Open

gemini-cli Bot added the status/pr-nudge-sent label Sep 26, 2026

This was referenced Sep 26, 2026

📊 AI CLI 工具社区动态日报 2026-09-26 96loveslife/big_model_radar#548

Open

📊 AI CLI 工具社区动态日报 2026-09-27 litang9/big_model_radar#458

Open

gemini-cli Bot closed this Oct 3, 2026

This was referenced Oct 3, 2026

📊 AI CLI 工具社区动态日报 2026-10-03 96loveslife/big_model_radar#580

Open

📊 AI CLI 工具社区动态日报 2026-10-04 litang9/big_model_radar#493

Open

Navigation Menu

Repository navigation

Fix/29365 duplicate tool responses - #29400

Fix/29365 duplicate tool responses#29400
abhashkumar9051 wants to merge 3 commits into
google-gemini:mainfrom
abhashkumar9051:fix/29365-duplicate-tool-responses

abhashkumar9051 commented Sep 18, 2026

Uh oh!

google-cla Bot commented Sep 18, 2026

Uh oh!

gemini-code-assist Bot commented Sep 18, 2026

Uh oh!

github-actions Bot commented Sep 18, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Uh oh!

gemini-cli Bot commented Sep 26, 2026

Uh oh!

gemini-cli Bot commented Oct 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

	it('should sync only the matching function response during sync', async () => {
	it('should preserve multi-modal sibling parts during sync', async () => {

Conversation

abhashkumar9051 commented Sep 18, 2026

Summary

Details

Related Issues

How to Validate

Pre-Merge Checklist

Uh oh!

google-cla Bot commented Sep 18, 2026

Uh oh!

gemini-code-assist Bot commented Sep 18, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

github-actions Bot commented Sep 18, 2026

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Sep 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-cli Bot commented Sep 26, 2026

Uh oh!

gemini-cli Bot commented Oct 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant